Jump to content
Sign in to follow this  
Suresh Jayaram

Is the data continuous or discrete?

Recommended Posts

FOLLOWING TOPIC PASTED FROM YAHOO GROUPS

ORIGINATOR: Rohit Bharadwaj

DATE: Mon Dec 8, 2008

Hello ,

Just 1 quick question ..! (Related to BPO Industry)

I get data from the client advisor wise in % with no base data .... Is that data continuous or discreet

Regards

Rohit

Share this post


Link to post
Share on other sites

This is a discrete data as to calculate % data we are counting good or bad out of total no. of opportunities. suppose we have 10 defects out of 100 opportunites then we are 90% accurate. Here to calculate accuracy we are counting the no. of opportunites and defects. So it is discete.

But I have alos seen % data being used as continuous also as it can be broken in to decimal points.

 

Share this post


Link to post
Share on other sites

I got an email from a past participant which I am posting here for everyone's benefit. Please feel free to post answers to this query.

Hope you are fine. I was a part of your December 2008 batch - Delhi.

I have a question and I am unsure of how to post it in the group.

If I am doing a green belt project to improve ASAT (Agent Satisfaction) scores of a process, what test should I use for Hypothesis Testing. According to my understanding, the 'Y' here is continuous data as the customers are rating the Agent on 3 parameters - Not met, Met and Exceeded. E.g:

Listening - Met, not met, exceeded

Knowledge - Met, not met, exceeded etc

The measure phase shows that First Time Resolution is the main reason why a customer is dissatisfied with the Agent. In this case, the 'X' is therefore discrete data.

So, how will I confirm that this 'X' that has been identified is the critical X?

Please advise.

Thanks,

Share this post


Link to post
Share on other sites
Guest V.B.MANIAN Manian

Hi,

  If we see the YX Matrix/Mapping as a tool in Measure phase, we can give some weighted scores to all the input causes for X cases & enumerate the table for the YX Matrix. We need to do this for all possible reasons it happens for that. Scoring more like FMEA type or how we do any customer survey in marketing / Customer satisfaction.  

We can also do a process FMEA here.

Then, it will be very clear as to which cause with a highest rating is our critical X to improve the agents rating. In hypothesis testing we should use Null hypothesis - use agent is meeting the requirement as one & another one as not meeting it. We need to use these type of weighted scores & along with random sampling find out the performance levels. Data(Y) is continuous only & measure looks like binomial yes/no or attribute. In the case of attribute data also we can convert them into variable by these creative methods. We can do F distribution variance if comparing with two agents or limited data. Then, T, normal, complete Z & so on. We need to look into the nature of the case. 

Note - In the case of Six sigma, y = f(x) is the transfer function. As per the changes in x or inputs, the output y varies. Continuous or discrete we refer to as how we plan to vary x. Accordingly y will also take that shape decided by the transfer function. In terms of variables we call y as dependent variable. Accordingly per the question, here we need to take X as discrete and Y as continuous, as already mentioned & explained therein.     

Thnx & rgds.

 

Share this post


Link to post
Share on other sites

In this exercise, the output Y is discrete because we have three possible outcomes (Met, Not Met, and Exceeded). The input FTR can also be discrete (Resolved First Time, and Not Resolved First Time).

The appropriate tool to use in this case would be a Chi-Square Test.

Randomly collect, say 100 samples for cases that were resolved the first time and for these cases, determine how many were rated "Met", "Not Met", and "Exceeded".

Similarly, randomly select 100 more samples for cases that were not resolved the first time and for these cases, determine how many were rated "Met", "Not Met", and Exceeded".

You can put your data in a two-way table (example follows):

post-63577-0-40338400-1318577676_thumb.j

If you perform a two-way table Chi-Square test, the P value is given as:

Chi-Sq = 1.425, DF = 2, P-Value = 0.490

Thus, for this example, since the P value is high, we accept the null hypothesis that there is no difference in the ratings for the agent whether they resolve the case FTR or not.

Thus, FTR is not a vital X for this example.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×